Connect with us

Published

on

Dado Ruvic | Reuters

Chinese artificial intelligence firm DeepSeek rocked markets this week with claims its new AI model outperforms OpenAI’s and cost a fraction of the price to build.

The assertions — specifically that DeepSeek’s large language model cost just $5.6 million to train — have sparked concerns over the eyewatering sums that tech giants are currently spending on computing infrastructure required to train and run advanced AI workloads.

But not everyone is convinced by DeepSeek’s claims.

CNBC asked industry experts for their views on DeepSeek, and how it actually compares to OpenAI, creator of viral chatbot ChatGPT which sparked the AI revolution.

What is DeepSeek?

Last week, DeepSeek released R1, its new reasoning model that rivals OpenAI’s o1. A reasoning model is a large language model that breaks prompts down into smaller pieces and considers multiple approaches before generating a response. It is designed to process complex problems in a similar way to humans.

DeepSeek was founded in 2023 by Liang Wenfeng, co-founder of AI-focused quantitative hedge fund High-Flyer, to focus on large language models and reaching artificial general intelligence, or AGI.

AGI as a concept loosely refers to the idea of an AI that equals or surpasses human intellect on a wide range of tasks.

Much of the technology behind R1 isn’t new. What is notable, however, is that DeepSeek is the first to deploy it in a high-performing AI model with — according to the company — considerable reductions in power requirements.

“The takeaway is that there are many possibilities to develop this industry. The high-end chip/capital intensive way is one technological approach,” said Xiaomeng Lu, director of Eurasia Group’s geo-technology practice.

“But DeepSeek proves we are still in the nascent stage of AI development and the path established by OpenAI may not be the only route to highly capable AI.” 

How is it different from OpenAI?

Read more DeepSeek coverage

In a technical report, the company said its V3 model had a training cost of only $5.6 million — a fraction of the billions of dollars that notable Western AI labs such as OpenAI and Anthropic have spent to train and run their foundational AI models. It isn’t yet clear how much DeepSeek costs to run, however.

If the training costs are accurate, though, it means the model was developed at a fraction of the cost of rival models by OpenAI, Anthropic, Google and others.

Daniel Newman, CEO of tech insight firm The Futurum Group, said these developments suggest “a massive breakthrough,” although he shed some doubt on the exact figures.

“I believe the breakthroughs of DeepSeek indicate a meaningful inflection for scaling laws and are a real necessity,” he said. “Having said that, there are still a lot of questions and uncertainties around the full picture of costs as it pertains to the development of DeepSeek.”

Meanwhile, Paul Triolio, senior VP for China and technology policy lead at advisory firm DGA Group, noted it was difficult to draw a direct comparison between DeepSeek’s model cost and that of major U.S. developers.

“The 5.6 million figure for DeepSeek V3 was just for one training run, and the company stressed that this did not represent the overall cost of R&D to develop the model,” he said. “The overall cost then was likely significantly higher, but still lower than the amount spent by major US AI companies.” 

DeepSeek wasn’t immediately available for comment when contacted by CNBC.

Comparing DeepSeek, OpenAI on price

DeepSeek and OpenAI both disclose pricing for their models’ computations on their websites.

DeepSeek says R1 costs 55 cents per 1 million tokens of inputs — “tokens” referring to each individual unit of text processed by the model — and $2.19 per 1 million tokens of output.

In comparison, OpenAI’s pricing page for o1 shows the firm charges $15 per 1 million input tokens and $60 per 1 million output tokens. For GPT-4o mini, OpenAI’s smaller, low-cost language model, the firm charges 15 cents per 1 million input tokens.

Skepticism over chips

LinkedIn co-founder Reid Hoffman: DeepSeek AI proves this is now a 'game-on competition' with China

Nvidia has since come out and said that the GPUs that DeepSeek used were fully export-compliant.

The real deal or not?

Industry experts seem to broadly agree that what DeepSeek has achieved is impressive, although some have urged skepticism over some of the Chinese company’s claims.

“DeepSeek is legitimately impressive, but the level of hysteria is an indictment of so many,” U.S. entrepreneur Palmer Luckey, who founded Oculus and Anduril wrote on X.

“The $5M number is bogus. It is pushed by a Chinese hedge fund to slow investment in American AI startups, service their own shorts against American titans like Nvidia, and hide sanction evasion.”

Seena Rejal, chief commercial officer of NetMind, a London-headquartered startup that offers access to DeepSeek’s AI models via a distributed GPU network, said he saw no reason not to believe DeepSeek.

“Even if it’s off by a certain factor, it still is coming in as greatly efficient,” Rejal told CNBC in a phone interview earlier this week. “The logic of what they’ve explained is very sensible.”

However, some have claimed DeepSeek’s technology might not have been built from scratch.

“DeepSeek makes the same mistakes O1 makes, a strong indication the technology was ripped off,” billionaire investor Vinod Khosla said on X, without giving more details.

It’s a claim that OpenAI itself has alluded to, telling CNBC in a statement Wednesday that it is reviewing reports DeepSeek may have “inappropriately” used output data from its models to develop their AI model, a method referred to as “distillation.”

“We take aggressive, proactive countermeasures to protect our technology and will continue working closely with the U.S. government to protect the most capable models being built here,” an OpenAI spokesperson told CNBC.

Commoditization of AI

However the scrutiny surrounding DeepSeek shakes out, AI scientists broadly agree it marks a positive step for the industry.

Yann LeCun, chief AI scientist at Meta, said that DeepSeek’s success represented a victory for open-source AI models, not necessarily a win for China over the U.S. Meta is behind a popular open-source AI model called Llama.

“To people who see the performance of DeepSeek and think: ‘China is surpassing the US in AI.’ You are reading this wrong. The correct reading is: ‘Open source models are surpassing proprietary ones’,” he said in a post on LinkedIn.

“DeepSeek has profited from open research and open source (e.g. PyTorch and Llama from Meta). They came up with new ideas and built them on top of other people’s work. Because their work is published and open source, everyone can profit from it. That is the power of open research and open source.”

WATCH: Why DeepSeek is putting America’s AI lead in jeopardy

Why China's DeepSeek is putting America's AI lead in jeopardy

– CNBC’s Katrina Bishop and Hayden Field contributed to this report

Continue Reading

Technology

UPS shares tank 17% after weak guidance, plan to slash Amazon deliveries by more than half

Published

on

By

UPS shares tank 17% after weak guidance, plan to slash Amazon deliveries by more than half

Amazon Prime and UPS trucks are seen on a building in Washington DC, United States on July 12, 2024. 

Jakub Porzycki | Nurphoto | Getty Images

Shares of United Parcel Service plunged more than 17% Thursday after the company issued weak revenue guidance for the year and said it planned to cut deliveries for Amazon, its largest customer, by more than half.

The shipping giant said in its fourth-quarter earnings report that it “reached an agreement in principle with its largest customer to lower its volume by more than 50% by the second half of 2026.”

At the same time, UPS said it’s reconfiguring its U.S. network and launching multi-year efficiency initiatives that it expects will result in savings of approximately $1 billion.

UPS CEO Carol Tome said on a call with investors that Amazon is UPS’ largest customer, but it’s not the company’s most profitable customer. “Its margin is very dilutive to the U.S. domestic business,” she added.

“We are making business and operational changes that, along with the foundational changes we’ve already made, will put us further down the path to become a more profitable, agile and differentiated UPS that is growing in the best parts of the market,” Tome said in a statement.

Read more CNBC tech news

Amazon spokesperson Kelly Nantel told CNBC in a statement that UPS had requested a reduction in volume “due to their operational needs.”

“We certainly respect their decision,” Nantel said in a statement. “We’ll continue to partner with them and many other carriers to serve our customers.”

Amazon said before the UPS announcement that it had offered to increase UPS’ volumes.

UPS forecast 2025 revenue of $89 billion, down from revenue of $91.1 billion in 2024. That’s well below consensus estimates for 2025 revenue of $94.88 billion, according to analysts polled by LSEG.

For the fourth quarter, UPS missed on revenue, reporting $25.30 billion versus $25.42 billion analysts anticipated in a survey by LSEG.

Amazon has long relied on a mix of major carriers for deliveries, including UPS, FedEx and the U.S. Postal Service. But it has decreased the number of packages sent through UPS and other carriers in recent years as it looks to have more control over deliveries.

Amazon has rapidly built up its own logistics empire since a 2013 holiday fiasco left its packages stranded in the hands of outside carriers. The company now oversees thousands of last-mile delivery companies that deliver packages exclusively for Amazon, as well as a budding in-house network of planes, trucks and ships. By some estimates, Amazon’s in-house logistics operations have grown to rival or exceed the size of major carriers.

UPS has, for its part, taken more aggressive cost-control measures, including catering to more profitable delivery customers. In recent quarters, UPS has benefited from an influx of volume from bargain retailers Temu and Shein, which have rapidly gained popularity in the U.S.

Last January, UPS laid off 12,000 employees as part of a bid to realize $1 billion in cost savings.

Continue Reading

Technology

Apple reports first-quarter earnings after the bell

Published

on

By

Apple reports first-quarter earnings after the bell

Apple CEO Tim Cook greets former President Barack Obama at the inauguration of U.S. President Donald Trump at the U.S. Capitol Rotunda in Washington, D.C., on Jan. 20, 2025.

Julia Demaree Nikhinson | Getty Images

Apple reports December-quarter earnings Thursday after the bell. 

The December quarter is Apple’s largest of the year, partially due to the holiday shopping season and also because it is the first full quarter of new iPhone sales.

While analysts are not worried about the company’s performance in the December quarter, many of them will look for what Apple signals about how its March quarter is shaking out.

Supply chain data points suggest Apple’s sales in China are weakening, and Apple Intelligence, the company’s suite of artificial intelligence features, is not available in Chinese yet.

“Specifically, iPhone 16 demand is not amplified by the introduction of iOS 18 and its Gen AI features. In fact, paradoxically, somehow demand is actually softer,” wrote Loop Capital analyst Ananda Baruah in a note earlier this month, downgrading Apple to hold. “We’re again looking for iPhone units to decline for the fourth consecutive year.”

Apple does not publish its unit sales, and does not give traditional guidance. New Chief Financial Officer Kevan Parekh, who assumed the role earlier this month, will likely give investors a few data points on Thursday’s call that analysts can use to estimate earnings per share and revenue for Apple’s March-quarter performance.

LSEG estimates Apple’s revenue will grow on an annual basis at about 3.8% to $124.13 billion. Apple said in October that it expected “low- to mid-single digit” sales growth during the quarter.

One of the biggest things analysts will be watching for is if Apple’s mainland China sales suggest that consumers in the country are shifting their preferences to locally made and designed devices.

“We believe that a major driver of growing competition within the smartphone market is due to growing preference for domestic brands within China,” wrote Goldman Sachs analyst Michael Ng in a Jan. 23 note.

One bright spot for Apple could be its services business, which includes products ranging from device warranties to the Apple TV+ streaming service. Barclays analysts said in a note earlier this month that services could grow as much as 14% on an annual basis, which could offset lower iPhone sales.

Apple is expected to be questioned over its plan for Trump’s proposed tariffs and its overall AI strategy.

Here is what to expect from Apple in the December quarter, per LSEG consensus estimates:

  • Earnings per share: $2.35
  • Revenue: $124.13 billion

Analysts are expecting guidance for the March quarter of $1.66 in earnings per share on $95.46 billion in revenue.

WATCH: Apple’s superficial problem is there’s not enough demand, says Jim Cramer

Apple's superficial problem is there's not enough demand, says Jim Cramer

Continue Reading

Technology

OpenAI partners with U.S. National Laboratories on scientific research, nuclear weapons security

Published

on

By

OpenAI partners with U.S. National Laboratories on scientific research, nuclear weapons security

OpenAI CEO Sam Altman speaks next to SoftBank CEO Masayoshi Son after U.S. President Donald Trump delivered remarks on AI infrastructure at the Roosevelt room at White House in Washington, U.S., January 21, 2025. 

Carlos Barria | Reuters

OpenAI on Thursday said the U.S. National Laboratories will be using its latest artificial intelligence models for scientific research and nuclear weapons security.

Under the agreement, up to 15,000 scientists working at the National Laboratories may be able to access OpenAI’s reasoning-focused o1 series. OpenAI will also work with Microsoft, its lead investor, to deploy one of its models on Venado, the supercomputer at Los Alamos National Laboratory, according to a release. Venado is powered by technology from Nvidia and Hewlett-Packard Enterprise.

OpenAI CEO Sam Altman announced the partnership at a company event called “Building to Win: AI Economics,” in Washington, D.C.

According to OpenAI, the new partnership will involve scientists using OpenAI’s technology to enhance cybersecurity to protect the U.S. power grid, identify new approaches to treating and preventing diseases and deepen understanding of fundamental mathematics and physics.

It will also involve work on nuclear weapons, “focused on reducing the risk of nuclear war and securing nuclear materials and weapons worldwide,” the company wrote. Some OpenAI researchers with security clearances will consult on the project.

Read more CNBC reporting on AI

Earlier this week, OpenAI released ChatGPT Gov, an AI platform built specifically for U.S. government use. OpenAI billed the new platform as a step beyond ChatGPT Enterprise as far as security. It will allow government agencies to feed “non-public, sensitive information” into OpenAI’s models while operating within their own secure hosting environments, the company said.

OpenAI said that since the beginning of 2024, more than 90,000 employees of federal, state and local governments have generated over 18 million prompts within ChatGPT, using the technology to translate and summarize documents, write and draft policy memos, generate code and build applications.

The government partnership follows a series of moves by Altman and OpenAI that appear to be targeted at appeasing President Donald Trump. Altman contributed $1 million to the inauguration, attended the event last week alongside other tech CEOs and recently signaled his admiration for the president.

Altman wrote on X that watching Trump “more carefully recently has really changed my perspective on him,” adding that “he will be incredible for the country in many ways.” OpenAI is also part of the recently announced Stargate project that involves billions of dollars in investment into U.S. AI infrastructure.

As OpenAI steps up its ties to the government, a Chinese rival is blowing up in the U.S. DeepSeek, an AI startup lab out of China, saw its app soar to the top of Apple’s App Store rankings this week and roiled U.S. markets on reports that its powerful model was trained at a fraction of the cost of U.S. competitors.

Altman described DeepSeek’s R1 model as “impressive,” and wrote on X that “we will obviously deliver much better models and also it’s legit invigorating to have a new competitor!”

WATCH: OpenAI highly overvalued

OpenAI is highly overvalued and DeepSeek just blew up their business model, says NYU's Gary Marcus

Continue Reading

Trending